Mammalian cortex operates in two fundamentally different modes. One, dominant during waking, is termed the upstate because of relatively high neuronal firing rates and synaptic activity. The other, oscillating with the upstate in the deepest stages of non-rapid eye-movement sleep, is characterized by a profound suppression of cell-firing and is termed the downstate. This slow oscillation (SO) has been intensively studied in animals with intracellular recordings, especially in model systems in vitro and in vivo under anesthesia. The basic phenomena have been reproduced from channel properties and synaptic connectivity in realistic Hodgkin-Huxley (H-H) computational models with limited numbers of cells4,6. Recent multi-microelectrode recordings in humans have demonstrated that the SO corresponds to .5-2Hz delta activity prominent in the stage 3 and 4 sleep EEG2, and further, that the downstate can occur in relative isolation as the K-Complex (KC) of stage 2 sleep1. These studies have established the basic local mechanisms of upstates and downstates, and their correspondence to prominent EEG phenomena that are easily observable in non-invasive recordings. However, important aspects of how they are triggered and synchronized remain unknown and controversial. Do SO and KC occur in all parts of the cortex? If so, do they preferentially occur in some areas? Do different SO and KC involve different cortical areas? Do they occur in all areas simultaneously or do they spread across the cortex? If they spread, is there a characteristic speed or point of origin? Do upstates and downstates differ in how they are triggered or synchronized? These are very complex questions regarding how billions of neurons are coordinated. Although empirical recordings are necessary to provide clues, these must be processed and interpreted with computational methods to make real headway. Biophysical and statistical forward and inverse computations are necessary to relate the microelectrode data to mesoscopic recordings (ECOG- electrocorticography) and non-invasive measures (MEG- magnetoencephalography and EEG). Neural modeling is necessary to test if specific hypothesized mechanisms for the origin and spread of the upstate and downstate correspond to the microscopic and mesoscopic recordings. Combined neural modeling and forward computations are needed to relate hypothesized mechanisms to EEG and MEG recordings. The proposed studies will yield a deep understanding of these fundamental states of the human cortex, computationally integrating animal with human recordings made at the channel, neuronal, circuit, system, and non-invasive whole-brain levels. Although the specific goal of this research proposal is to understand fundamental cortical functional states, further research based on the models could be applied to abnormal EEG/MEG from patients with sleep disorders, to predict the mechanisms that may be responsible for the observed abnormalities. The KC may function to prevent awakening; knowing its neural basis could lead to better treatment of insomnia. Most evidence suggests that the SO is the essential activity underlying the restorative processes of sleep. The SO also appears to play a central role in the consolidation of memories acquired in the preceding day. Sleep disorders have a causal relationship with reduced neurocognitive functions as well as variety of adverse physiologic and long-term health outcomes including all-cause mortality, diabetes, and cardiovascular disease. Over 30% of the general population complains about sleep-related problems. Sleep disorders - notably sleep apnea, sleep deprivation and sleepiness - affect 70 million Americans, resulting in $16 billion in annual healthcare expenses and $50 billion in lost productivity. In addition to significant economic benefits from healthcare, educational benefits include the training of graduate students and undergraduates who will be participating in the research. All of the software for running the models will be shared with other researchers and will be available through the internet in accordance with University policies. The new cross-disciplinary collaborations that will be established by the proposed research will lead to cross-disciplinary training of graduate students and postdoctoral fellows and will involve underrepresented groups and minorities. In addition to scientific presentations at meetings and lectures, the results of th research will be incorporated into teaching modules that could be used by K-12 teachers, in conjunction with the NSF sponsored Science of Learning Center at UCSD co-directed by Sejnowski. Intracranial recordings from humans are performed at Massachusetts General Hospital (MGH- Cash), New York Univ. (NYU- Thesen), Marseille (Chauvel), and Budapest (Ulbert). MEG/EEG recordings occur at UCSD (Halgren). Analysis and modeling occur at UC Riverside (UCR- Bazhenov), Paris (Destexhe), and UCSD (the central site- Halgren, Sejnowski, Dale and Hagler).